I've been working in the following problem but I can't fill in all the details.

Consider $E=\mathcal{C}\big([0, 1],\mathbb C \big)$ endowed with the norm induced by the product $\displaystyle\left\langle f,g\right\rangle =\int_0^1 f(t)\overline{g(t)}\,dt$. Fix $\alpha\in (0, 1)$ and define $u_\alpha\,\colon E\to \mathbb C$ by $$\displaystyle u_\alpha(f):= \int_0^\alpha t^2 f(t)\,dt.$$ I must prove that $u_\alpha\in E^*$, and also have to find its norm. Furthermore, I have to explain why $E$ is not a Hilbert space using this operator.

It is easy to see that $u_\alpha$ is a continuous linear functional $E\to \mathbb C$ because $|\phi_\alpha(f)|\leq \sqrt{\dfrac{\alpha^5}{5}}\, \lVert f \rVert_E$, and thus $ \lVert u_\alpha \rVert \leq\sqrt{\dfrac{\alpha^5}{5}}$, and I conjectured that this must be its norm, but I have failed trying to prove that (it may not be its norm in the first place, but I haven't got anything better)

For the second item, it suffices to show that the Riesz-Fréchet representation theorem doesn't hold for $u_\alpha$ (I've done it already, it is not hard at all though it is not evident)

One last question, for $n\in \mathbb N$, what is the norm of the following operator: $\displaystyle \phi_\alpha(f)= \int_0^\alpha t^n f(t)\,dt $?

Can you help me to find such norms? Thanks in advance!

  • $\begingroup$ How did you get the square root? (of $\alpha^5/5$) $\endgroup$ – Quoka Mar 6 '18 at 23:32
  • $\begingroup$ I just used the Cauchy-Schwarz inequality $\endgroup$ – EternalBlood Mar 6 '18 at 23:36
  • $\begingroup$ Using Cauchy-Scwartz I get $\alpha^5/5$, without the square root $\endgroup$ – Quoka Mar 6 '18 at 23:39
  • $\begingroup$ This is probably what you did: ${\displaystyle \left|\int _{\Omega}f(x){\overline {g(x)}}\,dx\right|^{2}\leq \int _{\Omega}|f(x)|^{2}\,dx\cdot \int _{\Omega}|g(x)|^{2}\,dx.}$ But one must extract square root to both sides $\endgroup$ – EternalBlood Mar 6 '18 at 23:59
  • 1
    $\begingroup$ Ok so to see that $\sqrt{\alpha^5/5}$ is actually the norm consider $f(t) = t^2\chi_{(0,a)}\sqrt{5/\alpha^5}$ $\endgroup$ – Quoka Mar 7 '18 at 0:09

The Cauchy-Schwarz inequality becomes equality iff one of the vectors is a multiple of the other, so take $f(t) = \sqrt{\frac{5}{\alpha^5}} t^2$. Then note that $\|f \|_2 = 1$ and $$\lvert u_\alpha(f) \rvert= \sqrt{\frac{5}{\alpha^5}}\int^\alpha_0 t^4 dt = \sqrt{\frac{\alpha^5}{5}}, $$ thus the equality is achieved and so $\|u_\alpha\| = \sqrt{\frac{\alpha^5}{5}}$.

The exact same reasoning works for $$\phi_\alpha(f) = \int^\alpha_0 t^n f(t) dt;$$ just use Cauchy-Schwarz and then choose $f$ so as to attain equality.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.